# Synth AI

Synth helps teams run Managed Research and Research Factory workflows on real
repositories and technical systems, with hosted workers, durable evidence,
reports, artifacts, usage controls, and operator review.

## Current Product Surfaces

- [Managed Research](https://docs.usesynth.ai/managed-research/intro):
  hosted AI research runs against repos and technical goals.
- [Managed Research Quickstart](https://docs.usesynth.ai/managed-research/quickstart):
  launch one hosted run and inspect evidence.
- [Managed Research MCP](https://docs.usesynth.ai/managed-research/mcp-quickstart):
  connect Codex or Claude Code to Synth's hosted MCP server.
- [Python SDK](https://docs.usesynth.ai/managed-research/sdk):
  use `synth-ai[research]` and `SynthClient().research` for programmatic runs.
- [Research Factory](https://docs.usesynth.ai/managed-research/factories):
  repeatable Managed Research programs built on the same SDK and MCP surfaces.
- [Runs and Evidence](https://docs.usesynth.ai/managed-research/runs-and-evidence):
  reports, artifacts, receipts, usage, and reviewable proof.

## Hosted MCP

- Server name: `synth-managed-research`
- Endpoint: `https://api.usesynth.ai/mcp`
- Docs: `https://docs.usesynth.ai/managed-research/mcp-quickstart`

## Optimizer Lane

- [Hosted Optimizers](https://docs.usesynth.ai/sdk/hosted-optimizers):
  hosted GEPA and GELO workflows with the `synth-optimizers` SDK and CLI.
- [Open-source GEPA](https://docs.usesynth.ai/sdk/oss-gepa):
  local GEPA examples and evidence.

## Canonical Packages

Use `synth-ai[research]` for Managed Research SDK and local MCP entrypoints.
Use `synth-optimizers` for GEPA and GELO optimizer workflows. Legacy package
names are compatibility surfaces only and should not be used for new
integrations.